cerevox 3.0.0-beta.25 → 3.0.0-beta.27

package/dist/mcp/servers/zerocut.js

@@ -513,7 +513,7 @@ server.registerTool('project-close', {
             .min(0)
             .max(20)
             .default(5)
-            .describe('Close the session after the specified number of minutes. Default is 5 minutes. 如果用户要求立即关闭会话，请将该参数设置为0！'),
+            .describe('Close the session after the specified number of minutes. Default is 5 minutes. 除非用户要求立即关闭会话，将该参数设置为0，否则应默认设为5'),
     },
 }, async ({ inMinutes }) => {
     try {
@@ -662,69 +662,6 @@ server.registerTool('retrieve-rules-context', {
         return createErrorResponse(`Failed to load rules context prompt for ${prompt}: ${error}`, 'retrieve-rules-context');
     }
 });
-// 列出项目下的所有文件
-server.registerTool('list-project-files', {
-    title: 'List Project Files',
-    description: 'List all files in the materials directory.',
-    inputSchema: {},
-}, async () => {
-    try {
-        // 验证session状态
-        const currentSession = await validateSession('list-project-files');
-        console.log('Listing project files...');
-        const terminal = currentSession.terminal;
-        if (!terminal) {
-            throw new Error('Terminal not available in current session');
-        }
-        let cwd;
-        try {
-            cwd = await terminal.getCwd();
-        }
-        catch (cwdError) {
-            console.error('Failed to get current working directory:', cwdError);
-            throw new Error('Failed to get current working directory');
-        }
-        console.log(`Current working directory: ${cwd}`);
-        // 安全地列出各目录文件，失败时返回空数组
-        const listFilesWithFallback = async (path, dirName) => {
-            try {
-                const files = await currentSession.files.listFiles(path);
-                console.log(`Found ${files?.length || 0} files in ${dirName}`);
-                return files || [];
-            }
-            catch (error) {
-                console.warn(`Failed to list files in ${dirName} (${path}):`, error);
-                return [];
-            }
-        };
-        const [rootFiles, materialsFiles, outputFiles] = await Promise.all([
-            listFilesWithFallback(cwd, 'root'),
-            listFilesWithFallback(`${cwd}/materials`, 'materials'),
-            listFilesWithFallback(`${cwd}/output`, 'output'),
-        ]);
-        const result = {
-            success: true,
-            cwd,
-            root: rootFiles,
-            materials: materialsFiles,
-            output: outputFiles,
-            totalFiles: rootFiles.length + materialsFiles.length + outputFiles.length,
-            timestamp: new Date().toISOString(),
-        };
-        console.log(`Total files found: ${result.totalFiles}`);
-        return {
-            content: [
-                {
-                    type: 'text',
-                    text: JSON.stringify(result),
-                },
-            ],
-        };
-    }
-    catch (error) {
-        return createErrorResponse(error, 'list-project-files');
-    }
-});
 server.registerTool('generate-character-image', {
     title: 'Generate Character Image',
     description: 'Generate a turnaround image or portrait for any character.',
@@ -931,49 +868,6 @@ ${roleDescriptionPrompt}
         return createErrorResponse(error, 'generate-character-image');
     }
 });
-server.registerTool('generate-line-sketch', {
-    title: 'Generate Line Sketch',
-    description: 'Generate line sketch material based on user prompt.',
-    inputSchema: {
-        prompt: zod_1.z.string().describe('The prompt to generate line sketch.'),
-        saveToFileName: zod_1.z
-            .string()
-            .describe('The filename to save the generated line sketch. 应该是png文件'),
-    },
-}, async ({ prompt, saveToFileName }) => {
-    try {
-        // 验证session状态
-        await validateSession('generate-line-sketch');
-        // 验证文件名
-        validateFileName(saveToFileName);
-        // 调用AI生成线稿
-        const res = await session.ai.generateLineSketch({ prompt });
-        if (res && res.url) {
-            // 保存到本地
-            await saveMaterial(session, res.url, saveToFileName);
-            const result = {
-                success: true,
-                url: res.url,
-                localPath: getMaterialUri(session, saveToFileName),
-                timestamp: new Date().toISOString(),
-            };
-            return {
-                content: [
-                    {
-                        type: 'text',
-                        text: JSON.stringify(result),
-                    },
-                ],
-            };
-        }
-        else {
-            throw new Error('No URL returned from AI service');
-        }
-    }
-    catch (error) {
-        return createErrorResponse(error, 'generate-line-sketch');
-    }
-});
 server.registerTool('upload-custom-material', {
     title: 'Upload Custom Material',
     description: 'Upload material files (images: jpeg/png, videos: mp4, audio: mp3) from the local filesystem to the materials directory. For video and audio files, duration information will be included in the response.',
@@ -1065,12 +959,18 @@ server.registerTool('generate-image', {
     description: `生成图片`,
     inputSchema: {
         type: zod_1.z
-            .enum(['banana', 'banana-pro', 'seedream', 'seedream-pro'])
+            .enum([
+            'banana',
+            'banana-pro',
+            'seedream',
+            'seedream-pro',
+            'line-sketch',
+        ])
             .optional()
             .default('seedream'),
         prompt: zod_1.z
             .string()
-            .describe('The prompt to generate. 一般要严格对应 storyboard 中当前场景的 start_frame 或 end_frame 中的字段描述'),
+            .describe('The prompt to generate. 一般要严格对应 storyboard 中当前场景的 start_frame 或 end_frame 中的字段描述，如果是生成线稿，则 type 使用 line-sketch'),
         sceneIndex: zod_1.z
             .number()
             .min(1)
@@ -2464,7 +2364,7 @@ server.registerTool('generate-scene-tts', {
             .describe('The volume of the tts.'),
         voiceID: zod_1.z
             .string()
-            .describe(`适合作为视频配音的音色ID，除非用户指定，否则你必须已通过 search_voice 工具检查确定该音色确实是存在的。`),
+            .describe(`适合作为视频配音的音色ID，除非用户指定，否则你必须确保已通过 pick-voice 工具挑选出真实存在的音色。`),
         context_texts: zod_1.z
             .array(zod_1.z.string())
             .default([])
@@ -2498,7 +2398,7 @@ server.registerTool('generate-scene-tts', {
             try {
                 const voice = (await ai.listVoices()).find(v => v.id === voiceID);
                 if (!voice) {
-                    return createErrorResponse(`Voice ${voiceID} not found in voice-list. Use search-voices tool to find available voices. 若用户坚持要使用该音色，需跳过一致性检查。`, 'generate-scene-tts');
+                    return createErrorResponse(`Voice ${voiceID} not found in voice-list. Use pick-voice tool to pick an available voice. 若用户坚持要使用该音色，需跳过一致性检查。`, 'generate-scene-tts');
                 }
                 const storyBoardPath = (0, node_path_1.resolve)(process.env.ZEROCUT_PROJECT_CWD || process.cwd(), projectLocalDir, storyBoardFile);
                 if ((0, node_fs_1.existsSync)(storyBoardPath)) {
@@ -2551,7 +2451,7 @@ server.registerTool('generate-scene-tts', {
         }
         console.log(`Generating TTS with voice: ${voiceID}, speed: ${finalSpeed}, text: ${text.substring(0, 100)}...`);
         if (voiceID.startsWith('BV0')) {
-            throw new Error(`BV0* 系列音色已弃用，你必须已通过 search_voice 工具检查确定该音色确实是存在的。`);
+            throw new Error(`BV0* 系列音色已弃用，你必须通过 pick-voice 工具挑选一个真实存在的音色。`);
         }
         const type = voiceID.startsWith('zh_') ||
             voiceID.startsWith('en_') ||
@@ -2706,7 +2606,7 @@ ${context_texts.join('\n')}
                         type: 'text',
                         text: JSON.stringify({
                             success: false,
-                            error: 'No TTS URL returned from AI service. You should use search-voices tool to find available voices.',
+                            error: 'No TTS URL returned from AI service. You should use pick-voice tool to pick an available voice.',
                             response: res,
                             timestamp: new Date().toISOString(),
                         }),
@@ -2939,67 +2839,71 @@ server.registerTool('get-schema', {
         return createErrorResponse(error, 'get-schema');
     }
 });
-server.registerTool('search-voices', {
-    title: 'Search Voices',
-    description: 'Search voices from doubao_voices_full based on languages, and gender. 搜索并选择符合要求的语音，在合适的情况下，优先采用 volcano_tts_2 类型的语音',
+server.registerTool('pick-voice', {
+    title: 'Pick Voice',
+    description: '根据用户需求，选择尽可能符合要求的语音，在合适的情况下，优先采用 volcano_tts_2 类型的语音',
     inputSchema: {
-        languages: zod_1.z
-            .array(zod_1.z.enum([
-            'zh',
-            'en',
-            'ja',
-            'ko',
-            'es',
-            'pt',
-            'nl',
-            'vi',
-            'ru',
-            'id',
-            'de',
-            'fr',
-            'it',
-            'ar',
-            'tr',
-            'uk',
-        ]))
+        prompt: zod_1.z
+            .string()
+            .describe('用户需求描述，例如：一个有亲和力的，适合给孩子讲故事的语音'),
+        custom_design: zod_1.z
+            .boolean()
             .optional()
-            .describe('Filter by languages (e.g., ["zh", "en"]). If not provided, no language filtering is applied.'),
-        gender: zod_1.z
-            .enum(['male', 'female'])
+            .describe('是否自定义语音，由于要消耗较多积分，因此**只有用户明确要求自己设计语音**，才将该参数设为true'),
+        custom_design_preview: zod_1.z
+            .string()
             .optional()
-            .describe('Filter by gender (male or female). If not provided, no gender filtering is applied.'),
+            .describe('用户自定义语音的预览文本，用于展示自定义语音的效果，只有 custom_design 为 true 时才需要'),
+        custom_design_save_to: zod_1.z
+            .string()
+            .optional()
+            .describe('自定义语音的保存路径，例如：custom_voice.mp3 custom_voice_{id}.mp3'),
     },
-}, async ({ languages, gender }) => {
+}, async ({ prompt, custom_design, custom_design_preview, custom_design_save_to, }) => {
     try {
         // 验证session状态
-        const currentSession = await validateSession('search-voices');
+        const currentSession = await validateSession('pick-voice');
         const ai = currentSession.ai;
-        let filteredVoices = await ai.listVoices();
-        // Filter by languages
-        if (languages && languages.length > 0) {
-            filteredVoices = filteredVoices.filter(voice => voice.languages &&
-                voice.languages.some((lang) => languages.includes(lang)));
-        }
-        // Filter by gender
-        if (gender) {
-            filteredVoices = filteredVoices.filter(voice => {
-                return voice.gender === gender;
+        if (custom_design) {
+            if (!custom_design_preview) {
+                throw new Error('custom_design_preview is required when custom_design is true');
+            }
+            const data = await currentSession.ai.voiceDesign({
+                prompt,
+                previewText: custom_design_preview,
             });
+            if (data.voice_id) {
+                const trial_audio = data.trial_audio;
+                let uri = '';
+                if (trial_audio) {
+                    uri = await saveMaterial(currentSession, trial_audio, custom_design_save_to || `custom_voice_${data.voice_id}.mp3`);
+                }
+                return {
+                    content: [
+                        {
+                            type: 'text',
+                            text: JSON.stringify({
+                                success: true,
+                                ...data,
+                                uri,
+                                timestamp: new Date().toISOString(),
+                            }),
+                        },
+                    ],
+                };
+            }
+            else {
+                throw new Error(`Voice design failed, ${JSON.stringify(data)}`);
+            }
         }
+        const data = await ai.pickVoice({ prompt });
         return {
             content: [
                 {
                     type: 'text',
                     text: JSON.stringify({
                         success: true,
-                        data: {
-                            totalCount: filteredVoices.length,
-                            voices: filteredVoices,
-                            filters: {
-                                languages: languages || null,
-                                gender: gender || null,
-                            },
-                        },
+                        ...data,
                         timestamp: new Date().toISOString(),
                     }),
                 },
@@ -3007,52 +2911,7 @@ server.registerTool('search-voices', {
         };
     }
     catch (error) {
-        return createErrorResponse(error, 'search-voices');
-    }
-});
-server.registerTool('voice-design', {
-    title: 'Voice Design',
-    description: 'Design a voice based on a prompt. The voice will be designed based on the prompt and preview text.',
-    inputSchema: {
-        prompt: zod_1.z.string().describe('The prompt to design the voice.'),
-        previewText: zod_1.z.string().describe('The preview text to design the voice.'),
-        saveToFileName: zod_1.z
-            .string()
-            .describe('The file name to save the designed voice. 应该是mp3文件'),
-    },
-}, async ({ prompt, previewText, saveToFileName }) => {
-    try {
-        const currentSession = await validateSession('voice-design');
-        const data = await currentSession.ai.voiceDesign({
-            prompt,
-            previewText,
-        });
-        if (data.voice_id) {
-            const trial_audio = data.trial_audio;
-            let uri = '';
-            if (trial_audio) {
-                uri = await saveMaterial(currentSession, trial_audio, saveToFileName);
-            }
-            return {
-                content: [
-                    {
-                        type: 'text',
-                        text: JSON.stringify({
-                            success: true,
-                            ...data,
-                            uri,
-                            timestamp: new Date().toISOString(),
-                        }),
-                    },
-                ],
-            };
-        }
-        else {
-            throw new Error(`Voice design failed, ${JSON.stringify(data)}`);
-        }
-    }
-    catch (error) {
-        return createErrorResponse(error, 'voice-design');
+        return createErrorResponse(error, 'pick-voice');
     }
 });
 server.registerTool('media-analyzer', {
@@ -3268,206 +3127,6 @@ server.registerTool('media-analyzer', {
         return createErrorResponse(error, 'media-analyzer');
     }
 });
-// server.registerTool(
-//   'image-aligner',
-//   {
-//     title: 'Image Aligner',
-//     description:
-//       'Analyze image quality and alignment with prompt using AI Image Quality Inspector.',
-//     inputSchema: {
-//       imageFileName: z
-//         .string()
-//         .describe('The image file name in materials directory to analyze.'),
-//       sceneIndex: z.number().min(1).describe('场景索引，从1开始的下标'),
-//       storyBoardFile: z
-//         .string()
-//         .optional()
-//         .default('storyboard.json')
-//         .describe('故事板文件路径'),
-//       imagePrompt: z
-//         .string()
-//         .optional()
-//         .describe('可选的图片提示词，如果提供则覆盖storyboard中的提示词'),
-//       customPrompt: z
-//         .string()
-//         .optional()
-//         .describe('可选的额外用户要求，用于补充图片质量评估的特定需求'),
-//     },
-//   },
-//   async ({
-//     imageFileName,
-//     sceneIndex,
-//     storyBoardFile = 'storyboard.json',
-//     imagePrompt,
-//     customPrompt,
-//   }) => {
-//     try {
-//       const currentSession = await validateSession('image-aligner');
-//       // 验证图片文件
-//       validateImageFile(imageFileName);
-//       // 获取图片 URL
-//       const imageUrl = getMaterialUri(currentSession, imageFileName);
-//       // 确定要使用的提示词
-//       let finalPrompt = imagePrompt;
-//       // 如果没有提供imagePrompt，则从storyboard中获取
-//       if (!imagePrompt) {
-//         try {
-//           const storyBoardPath = resolve(
-//             process.env.ZEROCUT_PROJECT_CWD || process.cwd(),
-//             projectLocalDir,
-//             storyBoardFile
-//           );
-//           if (existsSync(storyBoardPath)) {
-//             const storyBoardContent = await readFile(storyBoardPath, 'utf8');
-//             const storyBoard = JSON.parse(storyBoardContent);
-//             if (storyBoard.scenes && Array.isArray(storyBoard.scenes)) {
-//               const scene = storyBoard.scenes[sceneIndex - 1]; // sceneIndex 从1开始，数组从0开始
-//               if (scene) {
-//                 // 根据文件名判断优先级：若end_frame存在且imageFileName包含"_end"则优先取end_frame，否则取start_frame
-//                 if (scene.end_frame && imageFileName.includes('_end')) {
-//                   finalPrompt = scene.end_frame;
-//                 } else {
-//                   finalPrompt = scene.start_frame || scene.end_frame;
-//                 }
-//                 if (!finalPrompt) {
-//                   return createErrorResponse(
-//                     `场景 ${sceneIndex} 中未找到 start_frame 或 end_frame 提示词`,
-//                     'image-aligner'
-//                   );
-//                 }
-//               } else {
-//                 return createErrorResponse(
-//                   `在 ${storyBoardFile} 中未找到场景索引 ${sceneIndex}`,
-//                   'image-aligner'
-//                 );
-//               }
-//             } else {
-//               return createErrorResponse(
-//                 `${storyBoardFile} 文件格式不正确，缺少 scenes 数组`,
-//                 'image-aligner'
-//               );
-//             }
-//           } else {
-//             return createErrorResponse(
-//               `故事板文件不存在: ${storyBoardPath}`,
-//               'image-aligner'
-//             );
-//           }
-//         } catch (error) {
-//           return createErrorResponse(
-//             `读取或解析故事板文件失败: ${error}`,
-//             'image-aligner'
-//           );
-//         }
-//       }
-//       // 如果仍然没有提示词，返回错误
-//       if (!finalPrompt) {
-//         return createErrorResponse(
-//           '未提供 imagePrompt 且无法从故事板中获取提示词',
-//           'image-aligner'
-//         );
-//       }
-//       // 读取图片质量检查指南
-//       const alignerGuidelinePath = resolve(
-//         __dirname,
-//         './prompts/reasonings/image_aligner.md'
-//       );
-//       let alignerGuideline = '';
-//       try {
-//         alignerGuideline = await readFile(alignerGuidelinePath, 'utf8');
-//       } catch (error) {
-//         console.warn('无法读取图片质量检查指南:', error);
-//         alignerGuideline =
-//           '请对图片质量进行评估，包括构图、色彩、清晰度等方面。';
-//       }
-//       // 构建系统提示
-//       const systemPrompt = `你是一个专业的AI图片质量检查员。请根据以下指南对图片进行评估：
-// ${alignerGuideline}
-// 请严格按照指南中的JSON格式返回评估结果。`;
-//       // 构建用户提示
-//       const userPrompt = `请对这张图片进行质量评估。
-// 原始提示词：${finalPrompt}${
-//         customPrompt
-//           ? `
-// 额外要求：${customPrompt}`
-//           : ''
-//       }
-// 请按照指南要求，返回包含评分、问题列表和优化建议的JSON格式结果。`;
-//       // 调用AI模型进行图片质量评估
-//       const ai = currentSession.ai;
-//       const completion = await ai.getCompletions({
-//         model: 'Doubao-Seed-1.6',
-//         messages: [
-//           {
-//             role: 'system',
-//             content: systemPrompt,
-//           },
-//           {
-//             role: 'user',
-//             content: [
-//               {
-//                 type: 'image_url',
-//                 image_url: {
-//                   url: imageUrl,
-//                 },
-//               },
-//               {
-//                 type: 'text',
-//                 text: userPrompt,
-//               },
-//             ],
-//           },
-//         ],
-//       });
-//       const result = completion.choices[0]?.message?.content;
-//       if (!result) {
-//         throw new Error('No response from AI model');
-//       }
-//       // 解析AI响应
-//       let alignmentResult;
-//       try {
-//         // 尝试从响应中提取JSON
-//         const jsonMatch =
-//           result.match(/```json\s*([\s\S]*?)\s*```/) ||
-//           result.match(/\{[\s\S]*\}/);
-//         if (jsonMatch) {
-//           alignmentResult = JSON.parse(jsonMatch[1] || jsonMatch[0]);
-//         } else {
-//           // 如果没有找到JSON格式，尝试直接解析整个响应
-//           alignmentResult = JSON.parse(result);
-//         }
-//       } catch (error) {
-//         // 如果解析失败，返回原始响应
-//         alignmentResult = {
-//           error: 'JSON解析失败',
-//           raw_response: result,
-//         };
-//       }
-//       return {
-//         content: [
-//           {
-//             type: 'text',
-//             text: JSON.stringify({
-//               success: true,
-//               imageFileName,
-//               sceneIndex,
-//               storyBoardFile,
-//               imagePrompt: finalPrompt,
-//               customPrompt,
-//               promptSource: imagePrompt ? 'manual_override' : 'storyboard',
-//               analysis: alignmentResult,
-//               imageUrl,
-//               nextActionSuggest:
-//                 '可根据分析结果调整提示词，修改storyboard后，重新生成图片。',
-//             }),
-//           },
-//         ],
-//       };
-//     } catch (error) {
-//       return createErrorResponse(error, 'image-aligner');
-//     }
-//   }
-// );
 server.registerTool('audio-video-sync', {
     title: 'Audio Video Sync',
     description: 'Generate audio-video-synced video by matching video with audio. 还可以对口型。',
@@ -4399,6 +4058,69 @@ server.registerTool('search-context', {
         return createErrorResponse(error, 'search-context');
     }
 });
+// 列出项目下的所有文件
+server.registerTool('list-project-files', {
+    title: 'List Project Files',
+    description: 'List all files in the materials directory.',
+    inputSchema: {},
+}, async () => {
+    try {
+        // 验证session状态
+        const currentSession = await validateSession('list-project-files');
+        console.log('Listing project files...');
+        const terminal = currentSession.terminal;
+        if (!terminal) {
+            throw new Error('Terminal not available in current session');
+        }
+        let cwd;
+        try {
+            cwd = await terminal.getCwd();
+        }
+        catch (cwdError) {
+            console.error('Failed to get current working directory:', cwdError);
+            throw new Error('Failed to get current working directory');
+        }
+        console.log(`Current working directory: ${cwd}`);
+        // 安全地列出各目录文件，失败时返回空数组
+        const listFilesWithFallback = async (path, dirName) => {
+            try {
+                const files = await currentSession.files.listFiles(path);
+                console.log(`Found ${files?.length || 0} files in ${dirName}`);
+                return files || [];
+            }
+            catch (error) {
+                console.warn(`Failed to list files in ${dirName} (${path}):`, error);
+                return [];
+            }
+        };
+        const [rootFiles, materialsFiles, outputFiles] = await Promise.all([
+            listFilesWithFallback(cwd, 'root'),
+            listFilesWithFallback(`${cwd}/materials`, 'materials'),
+            listFilesWithFallback(`${cwd}/output`, 'output'),
+        ]);
+        const result = {
+            success: true,
+            cwd,
+            root: rootFiles,
+            materials: materialsFiles,
+            output: outputFiles,
+            totalFiles: rootFiles.length + materialsFiles.length + outputFiles.length,
+            timestamp: new Date().toISOString(),
+        };
+        console.log(`Total files found: ${result.totalFiles}`);
+        return {
+            content: [
+                {
+                    type: 'text',
+                    text: JSON.stringify(result),
+                },
+            ],
+        };
+    }
+    catch (error) {
+        return createErrorResponse(error, 'list-project-files');
+    }
+});
 server.registerTool('build-capcat-draft', {
     title: 'Build CapCut Draft',
     description: 'Read draft_content.json file, parse JSON and generate URIs for all assets in timeline tracks, then output the processed JSON string.',